The best reply to a customer during an outage is a short one, sent early, by someone who can see the whole picture: the check that failed at 07:12, the parts of the product it touches, what the status page already says, and the similar thing this customer raised in March. It takes a minute to write, and nobody had to go and ask anyone.
That reply is harder to write when those details live in separate products. Support tooling gets bought one purchase at a time, each for a sound reason, and each product keeps its own record of what is happening. What follows is one incident traced through a common stack: where the customer's context is dropped, and what the stack costs.
The stack we'll follow
Take a B2B SaaS company running Help Scout for the shared inbox and help centre, Atlassian Statuspage for the public status page, PagerDuty for on-call, and UptimeRobot for checks. Each of those is a good product, bought to solve the problem in front of the company that week, and nothing that follows is a criticism of any of them. What matters is what happens to a customer's problem as it crosses between them.
One incident, followed through the hand-offs
It is 07:12 on a Tuesday and the API has started returning errors.
The check fails. UptimeRobot fires an alert within its check interval, and its PagerDuty integration, which unlocks on the Team plan, carries the monitor's name, the URL and the status code onward. It cannot carry anything about who is affected, because a monitor holds no record of the customers.
The alert becomes an incident. PagerDuty pages whoever is on call, records the acknowledgement at 07:14 and starts a timeline. That timeline is about the alert and the responder, which is what an on-call product is for. The customer does not appear in it.
Someone updates the status page. Statuspage has its own login, so a human has to go there and write a post, usually one of the people fixing the thing. The subscribers who receive it are a list Statuspage holds, separate from the customer contacts in Help Scout even though they are largely the same people.
The ticket arrives. At 07:31 a customer emails. Help Scout has their whole history and nothing about the incident, so whoever picks it up has to find out what is happening, then restate it by hand, and again for the next ticket. When the incident resolves at 08:05, none of those tickets change.
The integrations between these products do work. What they cannot do is give them one shared record of the problem, so each keeps its own version, and the person replying is the one who reconciles them by hand, in the half hour when there is least time for it.
On any single day the cost of that is small: a reply at 07:50 instead of 07:35, a status page that says "investigating" for forty minutes after the fix, a follow-up that never gets sent. Customers form their view of a company from them cumulatively.
What this stack costs
Buying one product at a time shows up in the bill too. The figures below come from our support stack cost calculator, which picks the cheapest published plan that fits the numbers you give it, free tiers included. They are on monthly billing, converted from US dollars at £0.78 to the dollar, before VAT, and were checked on 3 September 2026. Treat them as a floor rather than a quote.
The example holds 250 status page subscribers and 15 monitors steady, and varies the two things that move the bill: the number of people on the team and the number of customer contacts they support.
| Scenario | Help Scout | Statuspage | PagerDuty | UptimeRobot | Stack per month |
|---|---|---|---|---|---|
| 5 people, 100 contacts | Free | Hobby | Free | Team, plus two extra login seats | about £88 |
| 5 people, 250 contacts | Standard, five seats | Hobby | Free | Team, plus two extra login seats | about £205 |
| 6 people, 250 contacts | Standard, six seats | Startup | Professional, six users | Team, plus three extra login seats | about £415 |
At five people and 250 contacts the stack comes to about £205 a month. Help Scout's free tier stops at 100 contacts helped in a month, so the company is on Standard at a price per seat. Statuspage needs the Hobby tier for 250 subscribers and a custom domain. PagerDuty is free for up to five users. UptimeRobot's Team plan includes three logins, so the other two people are paid seats.
Now hire one more person. The same company at six people is about £415 a month, and every part of that jump is a headcount cap rather than a customer number. PagerDuty's free tier ends at five users, so everyone moves onto Professional at a price per user, which is the single biggest step. Statuspage's Hobby tier allows five team members, so the page has to move to Startup. Help Scout and UptimeRobot each add a seat.
The smaller end of the table deserves the same honesty. At five people and 100 contacts the same stack costs about £88, because Help Scout's and PagerDuty's free tiers both fit. Free tiers are genuinely good at small numbers. What adds cost is the per-seat and per-team-member meters in the stack, which tick when a colleague is added whether or not there are more customers to support.
How many tools does a small SaaS need for support?
There is no correct number, and the counts that get quoted are for whole companies rather than support teams. Zylo's 2026 SaaS Management Index reports that organisations "leave an average of 36% of their SaaS licenses unused" when measured against industry-recommended utilisation levels (2026 SaaS Management Index, 29 January 2026). The buying keeps running ahead of the using.
For support, the more useful question is how many places one customer's problem has to be written down before it is answered. A stack is the right size when the person replying can see the failed check, the incident, what has been published, and the customer's own history from the screen they reply on.
What "unified around the customer" means in practice
A stack is unified around the customer when the customer is the record everything else attaches to, rather than something each product keeps a partial copy of. Two concrete properties follow.
The whole team can be in the tool, including the engineer on call and the founder who still answers tickets on a Friday, because a bill that follows customer contacts does not go up when a colleague is added. And the record persists: when the same customer writes in three months later, the person replying sees the earlier incident and the earlier conversation, and answers with both in front of them.
None of this requires a single product; a team could wire its existing tools into something close to it. The Tuesday morning is a reasonable test of whether the shared record is there.
Where StayUpfront fits
StayUpfront is built on that shared record. Web checks raise incidents automatically, incident updates go to the status page and its subscribers as part of writing them, support tickets link to the incidents they are about, and Una, the AI in StayUpfront, suggests the link and drafts a reply for a person to review and send. On-call schedules, escalation paths, the customer's history and the docs sit alongside.
It is priced by the customer contacts you support, with the whole team included. The example company, at 250 contacts, would be on Growth at £149 a month at the standing price, against about £205 for the stack at five people and about £415 at six. At 100 contacts it would be on Starter at £79, against the stack's £88, which is the honest size of the gap while free tiers are doing the work. Early customers are on founding prices, which put Growth at £99 and Starter at £49. All figures ex-VAT, and the pricing page has the rest of the plans.
To check the maths against your own numbers, the calculator shows every vendor's plan and why it was chosen. To see the Tuesday morning run end to end, there is a 14-day free trial with nothing charged until it ends.